One issue I could see on AIX 5.3: At one of my customer they have got the ThreadLimit of 2500 set on web server in httpd.conf file.
Currently 2000 users have logged in each attaing a single instance of httpd.
While in ps -ef for httpd process it is showing thread count (thcount) almost 3000. This should have shown atmost 2500 (or infact would have matched the number of users logged into i.e almost 2000)
Why the difference?
check for close_wait or fin_wait_2 connections
netstat -a | grep -iE "(close_wait|fin_wait_2)"
maybe the tcp connections do not shut down correctly, and the process remains until a timeout is reached
There does exisit these waits. What setting to be lookedout for streamlining these waits?
netstat -an | awk '{ print
$NF}' | sort | uniq -c | awk
'{ if ( $1> 10) print }'
13 0
1392 ESTABLISHED
39 FIN_WAIT_1
1755 FIN_WAIT_2
12 SYN_RCVD
352 TIME_WAIT